From d4e16e4b0eb59d37ed7bc6171af8aa230c6dc22d Mon Sep 17 00:00:00 2001 From: Emmanuel Fromager Date: Wed, 12 Feb 2020 12:04:03 +0100 Subject: [PATCH] Manu: saving work --- Manuscript/eDFT.tex | 67 +++++++++++++++++++++++++++++---------------- 1 file changed, 44 insertions(+), 23 deletions(-) diff --git a/Manuscript/eDFT.tex b/Manuscript/eDFT.tex index 6b62464..7a2516b 100644 --- a/Manuscript/eDFT.tex +++ b/Manuscript/eDFT.tex @@ -379,8 +379,8 @@ c}\left[n\right]}{\partial w_K}\right|_{n=n_{\bmg^{\bw}}} , \eeq where ${\bm -h}\equiv\langle\AO{\mu}\vert-\frac{1}{2}\nabla_{\br}^2+v_{\rm -ne}(\br)\vert\AO{\nu}\rangle$ and ${\bm G}\equiv{\bm J}-{\bm K}$ denote +h}\equiv\left\{\langle\AO{\mu}\vert-\frac{1}{2}\nabla_{\br}^2+v_{\rm +ne}(\br)\vert\AO{\nu}\rangle\right\}_{\mu\nu}$ and ${\bm G}\equiv{\bm J}-{\bm K}$ denote the Coulomb-exchange integrals. %%%%%%%%%%%%%%%%%%%%% @@ -419,6 +419,46 @@ w}_K \subsection{Approximations} +As Hartree and exchange energies cannot be separated in the +one-dimension systems considered in the rest of this work, we will substitute the Hartree--Fock +density-matrix-functional interaction energy, +\beq\label{eq:eHF-dens_mat_func} +W_{\rm +HF}\left[{\bmg}\right]=\frac{1}{2} \Tr(\bmg \, \bG \, \bmg), +\eeq +for the Hx density-functional energy in the variational energy +expression of Eq.~(\ref{eq:var_ener_gokdft}): +\beq +{\bmg}^{\bw}\approx\argmin_{{\bm\gamma}^{\bw}} +\Big\{ +{\rm +Tr}\left[{\bm \gamma}^{{\bw}}{\bm h}\right]+W_{\rm +HF}\left[{\bm\gamma}^{\bw}\right] ++ +{E}^{{\bw}}_{\rm +c}\left[n_{\bm\gamma^{\bw}}\right] +%+E^{{\bw}}_{\rm c}\left[n_{\hat{\Gamma}^{{\bw}}}\right] +\Big\}. +\nonumber\\ +\eeq +Note that this approximation, where the ensemble density matrix is +optimized from a non-local exchange potential [rather than a local one, +as expected from Eq.~(\ref{eq:var_ener_gokdft})] is applicable to real +(three-dimension) systems. As readily seen from +Eq.~(\ref{eq:eHF-dens_mat_func}), {\it ghost-interaction} errors will be +introduced in the ensemble HF interaction energy: + + + +In order to remove ghost interactions from the variational energy +expression used in the first step, we then employ the (in-principle-exact) +expression in Eq.~(\ref{eq:exact_ind_ener_OEP-like}). In this second +step, the response of the individual density matrices to weight +variations (last term on the right-hand side of +Eq.~(\ref{eq:exact_ind_ener_OEP-like})) is neglected. The complete GIC +procedure can be summarized as follows, +and + In order to compute (approximate) energy levels within generalized GOK-DFT we use a two-step procedure. The first step consists in optimizing variationally the ensemble density matrix according to @@ -438,27 +478,8 @@ c}[n]=\int d\br\;n(\br)\epsilon_{c}^{\bw}(n(\br)). \eeq More details about the construction of such a functional will be given in the -following. In order to remove ghost interactions from the variational energy -expression used in the first step, we then employ the (in-principle-exact) -expression in Eq.~(\ref{eq:exact_ind_ener_OEP-like}). In this second -step, the response of the individual density matrices to weight -variations (last term on the right-hand side of -Eq.~(\ref{eq:exact_ind_ener_OEP-like})) is neglected. The complete GIC -procedure can be summarized as follows, -\beq -{\bmg}^{\bw}\approx\argmin_{{\bm\gamma}^{\bw}} -\Big\{ -{\rm -Tr}\left[{\bm \gamma}^{{\bw}}{\bm h}\right]+W_{\rm -HF}\left[{\bm\gamma}^{\bw}\right] -+ -{E}^{{\bw}}_{\rm -c}\left[n_{\bm\gamma^{\bw}}\right] -%+E^{{\bw}}_{\rm c}\left[n_{\hat{\Gamma}^{{\bw}}}\right] -\Big\}, -\nonumber\\ -\eeq -and +following. + \beq E^{(I)}&&\approx{\rm Tr}\left[{\bmg}^{(I)}{\bm h}\right]